Contemporary Linguistics: An Introduction, 5th edition, Chapter 4: Morphophonemics, 1
Morphophonemics
As noted in the text, a word’s pronunciation can be sensitive to morphological
factors. Morphophonemic variation is characterized by two signposts—it tends
to occur at morpheme boundaries and it involves sounds that are associated
with separate phonemes. The study of these phenomena is known as
morphophonemics. We will consider two morphophonemic phenomena here,
one involving the selection of the correct form of the English plural suffix
(briefly mentioned in the textbook) and the other involving plurals in which the
final consonant of the root changes (as in thief-thieves).
Deriving allomorphs
When linguists find that the same morpheme has more than one form (or
allomorph), they often seek to set up a single underlying representation for the
morpheme and to formulate rules that will derive the appropriate pronunciation
for any particular context. To see how this works, let us consider in more detail
the precise environments in which the different allomorphs of the English plural
morpheme occur (see Table 1, which follows). To keep matters as simple as
possible, we will use phonemic transcription here rather than phonetic
transcription. (Recall that the alternation in question involves sounds that
belong to separate phonemes.)
Contemporary Linguistics: An Introduction, 5th edition, Chapter 4: Morphophonemics, 2
As you can see, the choice of plural allomorph is determined by the final
segment of the base. Bases that end in a nonstrident voiceless consonant take
the voiceless /-s/ allomorph. Bases that end in a vowel or a nonstrident voiced
consonant take the voiced /-z/ allomorph. And bases that end in a strident
(sibilant) consonant occur with the /-əz/ allomorph, to avoid having an illegal
consonant cluster, as we will see.
In selecting the underlying representation of a morpheme, it is common
to choose the allomorph with the widest distribution. This happens to be /-z/ in
the case of the English plural morpheme because it occurs after most voiced
consonants and after all vowels (which are also voiced). This is no accident, of
course, since /-z/ too is voiced.
The words lips, pills, and glasses therefore have the underlying
representations depicted in 1).
Contemporary Linguistics: An Introduction, 5th edition, Chapter 4: Morphophonemics, 3
1)
lɪp-z ‘lips’
pɪl-z ‘pills’
glæs-z ‘glasses’
Two rules are now required to derive the correct final form of these
words. The first rule, which we will call Coda Epenthesis, inserts a schwa
whenever the /-z/ appears after a base that ends in a strident consonant,
thereby breaking up the otherwise illegal consonant cluster.
2)
Coda Epenthesis
Ø →
ə
/[+strident] ______ -z
The second rule needed to ensure that the plural morpheme has the right
form devoices /-z/ when it occurs after a voiceless consonant in the same coda.
3)
Devoicing
-z
→
[-voice] / [-voice] ______ ]σ
Like Coda Epenthesis, Devoicing helps ensure that English plurals comply
with the usual phonotactic constraints of the language: no English words
contain a coda in which a voiceless consonant is followed by a voiced one.
Figure 1 illustrates the derivations of all three plural allomorphs.
Notice that the ordering of the rules is crucial here. If Devoicing applied first, as
shown in Figure 2, the plural ending of forms such as glasses would incorrectly
end up as */glæs-s/, since Coda Epenthesis would not apply.
Contemporary Linguistics: An Introduction, 5th edition, Chapter 4: Morphophonemics, 4
Conditioning by morphological class
A second type of morphophonemic phenomenon is manifested in just a
subclass of morphemes. This too can be illustrated with the help of English
plurals.
As the examples in Table 2 show, English includes a limited class of
words in which an /f/ at the end of the base alternates with /v/ when there is a
plural suffix.
The alternating class is unproductive, and new words with final /f/ entering
English do not exhibit the f-v alternation. That is why a hypothetical new word
such as nif would have /nɪfs/ rather than */nɪvz/ as its plural form.
We can formulate the process that is responsible for the f-v alternation as
follows.
4)
/f/ Voicing
f → v / ___ ]-Plural
Notice the usual hallmarks of a morphophonemic phenomenon. The
alternation takes place at a morpheme boundary, and it involves sounds that
belong to separate phonemes (/f/ and /v/). Moreover, the phenomenon affects
only certain root morphemes (e.g. thief, but not chief) and only the plural suffix
(there is no f-v alternation with the possessive morpheme -’s, which is why we
say his wife’/s/ car, not *his wive-’/z/ car). Morphophonemic rules are thus
very different from allophonic rules, which apply very generally—all vowels
occurring before a nasal in the same syllable are nasalized; all vowels occurring
before a voiced consonant in the same syllable are lengthened; and so on. But
only some base-final /f/s are converted to /v/ in the plural.
Figure 3 illustrates how the plural forms of thief and chief are derived
from the appropriate underlying representations.
Contemporary Linguistics: An Introduction, 5th edition, Chapter 4: Morphophonemics, 5
Once again, rule ordering is important here. Unless the /f/ Voicing rule
applies before Devoicing, the final form of thieves will be */θiv-s/, with the /-s/
allomorph of the plural, as shown in Figure 4.
Other examples of underlying representations
As the preceding examples help show, the underlying representations
needed to account for morphophonemic phenomena may be quite unlike a
word’s phonetic form. Another example of this involves the morpheme electric
in words like electrical and electricity, for which linguists often posit the
underlying representations in 5).
5)
/ilɛktrɪk-l /
/ilɛktrɪk-ɪti/
A morphophonemic rule then converts the /k/ of the base to /s/ in front of the
affix -ity. Notice once again that the rule involves sounds that belong to
separate phonemes (/k/ and /s/) and that it applies only at a morpheme
boundary, which is why we don’t pronounce kill as sill!
The variant of the base that ends in k is chosen as underlying for two
reasons. First, the base electri/k/ has a wider distribution than the allomorph
electri/s/, since it occurs in words such as electrical as well as in the unsuffixed
form electric. Second, a rule that fronts a /k/ to /s/ before the high front vowel
of the suffix is phonetically more natural than a rule that changes an /s/ to /k/
in final position or before the suffix -al. (Recall that /s/ is produced in the front
portion of the mouth with the tongue tip raised.)
A more complicated example, shown in Figure 5, involves the three-way
alternation involving submit, submissive, and submission. The underlying
representation for the root morpheme is /sʌbmɪt/, with a /t/-to-/s/ rule
Contemporary Linguistics: An Introduction, 5th edition, Chapter 4: Morphophonemics, 6
accounting for the /s/ before the suffix –ive in the second form and a /t/-to-/ʃ/
rule giving the /ʃ/ before the suffix –ion in the third word.
This sort of analysis has proven to be very popular over the years,
primarily because it allows linguists to posit a single underlying representation
for each morpheme and to find general rules that predict its actual
pronunciation in particular contexts.